Amazon AWS Certified Machine Learning Engineer - Associate MLA-C01 Practice Test 7

100% Latest & Updated Amazon AWS Certified Machine Learning Engineer - Associate MLA-C01 Practice Test Questions, Exam Dumps & Verified Answers!
30 Days Free Updates, Instant Download!

Question 60

A single ML platform team is handling two distinct changes for the same system. Workstream A: A risk rating has meaningful ordered tiers low, medium, and high, and a product scoring model should retain that order in a compact feature. Workstream B: Analysts need to compare transformations, visualize feature distributions, and export reproducible SageMaker processing code without maintaining a bespoke notebook transformation library. Select exactly TWO measures, one that directly resolves each workstream.

Set up a Ground Truth labeling job with quality controls and appropriate workers.
Encode the ordered tiers using a deliberate ordinal mapping and handle unknown tiers explicitly.
Create a versioned Data Wrangler flow and export its supported transformations into the ML pipeline.
Perform timestamp-aware historical feature joins using available offline Feature Store records.
Implement the join in a correctly partitioned AWS Glue Spark ETL job.

Explanation

Correct Answers

This action fits the situation: A risk rating has meaningful ordered tiers low, medium, and high, and a product scoring model should retain that order in a compact feature. It implements the needed technical control: Encode the ordered tiers using a deliberate ordinal mapping and handle unknown tiers explicitly. The intended outcome is that the numeric feature preserves the intended direction and relative rank of categories. The important limitation is that an ordinal code assumes meaningful order and is wrong for unordered product labels.

This action fits the situation: Analysts need to compare transformations, visualize feature distributions, and export reproducible SageMaker processing code without maintaining a bespoke notebook transformation library. It implements the needed technical control: Create a versioned Data Wrangler flow and export its supported transformations into the ML pipeline. The intended outcome is that exploratory feature preparation becomes repeatable in the managed workflow. The important limitation is that interactive exploration alone does not replace production input validation or version control.

Incorrect Answers

That alternative addresses a different task: A retailer with an existing SageMaker Ground Truth account has thousands of ambiguous product images and needs human labels with periodic consensus verification for supervised classification. The selected workloads instead require Ordinal feature handling, Data Wrangler visual pipeline.

That alternative addresses a different task: A model that predicts defaults at loan origination currently joins the customer's current account balance rather than the balance known on the origination date. The selected workloads instead require Ordinal feature handling, Data Wrangler visual pipeline.

That alternative addresses a different task: A 4 TB training dataset needs joins across multiple S3 partitions, and a single-node pandas job is exceeding memory. The selected workloads instead require Ordinal feature handling, Data Wrangler visual pipeline.

Question 61

A single ML platform team is handling two distinct changes for the same system. Workstream A: Transaction dollar amounts span several orders of magnitude, and a linear model is dominated by the largest outliers despite consistent labels. Workstream B: A data operations analyst cleans inconsistent dates and malformed postal codes in a structured dataset using an auditable sequence of low-code rules. Select exactly TWO measures, one that directly resolves each workstream.

Set up a Ground Truth labeling job with quality controls and appropriate workers.
Perform timestamp-aware historical feature joins using available offline Feature Store records.
Create and execute a versioned AWS Glue DataBrew recipe on the incoming dataset.
Apply an appropriate log1p-style transform to nonnegative transaction amounts before training.
Fit or choose a supported text tokenizer and reuse its saved vocabulary and preprocessing settings for inference.

Explanation

Correct Answers

This action fits the situation: A data operations analyst cleans inconsistent dates and malformed postal codes in a structured dataset using an auditable sequence of low-code rules. It implements the needed technical control: Create and execute a versioned AWS Glue DataBrew recipe on the incoming dataset. The intended outcome is that repeat data cleaning follows documented transform steps with reviewable outputs. The important limitation is that a recipe designed on a small sample must be tested against unusual production values.

This action fits the situation: Transaction dollar amounts span several orders of magnitude, and a linear model is dominated by the largest outliers despite consistent labels. It implements the needed technical control: Apply an appropriate log1p-style transform to nonnegative transaction amounts before training. The intended outcome is that the feature's long tail becomes less extreme, improving numerical modeling stability. The important limitation is that a log transform does not automatically fix mislabeled records and requires care with negative values.

Incorrect Answers

That alternative addresses a different task: A retailer with an existing SageMaker Ground Truth account has thousands of ambiguous product images and needs human labels with periodic consensus verification for supervised classification. The selected workloads instead require Heavy tailed feature transform, Repeatable DataBrew recipe.

That alternative addresses a different task: A model that predicts defaults at loan origination currently joins the customer's current account balance rather than the balance known on the origination date. The selected workloads instead require Heavy tailed feature transform, Repeatable DataBrew recipe.

That alternative addresses a different task: A service classification model receives free-text customer tickets and must convert words into consistent model input tokens while preserving the same vocabulary during inference. The selected workloads instead require Heavy tailed feature transform, Repeatable DataBrew recipe.

Question 62

A team is choosing a specific control for this workload: A credit risk team wants a smooth age measurement grouped into stable policy bands rather than a continuously varying predictor for a regulated report. Which action best addresses the actual requirement?

Create explicit numeric bins with documented boundaries and a consistent unseen-value policy.
Apply an appropriate log1p-style transform to nonnegative transaction amounts before training.
Perform timestamp-aware historical feature joins using available offline Feature Store records.
Fit the standardization parameters on training data only and apply the frozen transformation to validation and test splits.

Explanation

Correct Answer

This action fits the situation: A credit risk team wants a smooth age measurement grouped into stable policy bands rather than a continuously varying predictor for a regulated report. It implements the needed technical control: Create explicit numeric bins with documented boundaries and a consistent unseen-value policy. The intended outcome is that risk decisions can use stable and interpretable age groups. The important limitation is that binning discards within-bin detail and can amplify fairness concerns around thresholds.

Incorrect Answers

This action is appropriate for another requirement: Transaction dollar amounts span several orders of magnitude, and a linear model is dominated by the largest outliers despite consistent labels. It does not solve the presented constraint: A credit risk team wants a smooth age measurement grouped into stable policy bands rather than a continuously varying predictor for a regulated report. The matching action for this case is: Create explicit numeric bins with documented boundaries and a consistent unseen-value policy.

This action is appropriate for another requirement: A model that predicts defaults at loan origination currently joins the customer's current account balance rather than the balance known on the origination date. It does not solve the presented constraint: A credit risk team wants a smooth age measurement grouped into stable policy bands rather than a continuously varying predictor for a regulated report. The matching action for this case is: Create explicit numeric bins with documented boundaries and a consistent unseen-value policy.

This action is appropriate for another requirement: A churn model uses numerical features with different units, but the team accidentally fits a scaling transform on the complete dataset before holding out validation records. It does not solve the presented constraint: A credit risk team wants a smooth age measurement grouped into stable policy bands rather than a continuously varying predictor for a regulated report. The matching action for this case is: Create explicit numeric bins with documented boundaries and a consistent unseen-value policy.

Question 63

An ML engineering team faces this production constraint: A 4 TB training dataset needs joins across multiple S3 partitions, and a single-node pandas job is exceeding memory. Which action best addresses the actual requirement?

Apply an appropriate log1p-style transform to nonnegative transaction amounts before training.
Create a versioned Data Wrangler flow and export its supported transformations into the ML pipeline.
Fit or choose a supported text tokenizer and reuse its saved vocabulary and preprocessing settings for inference.
Implement the join in a correctly partitioned AWS Glue Spark ETL job.

Explanation

Correct Answer

This action fits the situation: A 4 TB training dataset needs joins across multiple S3 partitions, and a single-node pandas job is exceeding memory. It implements the needed technical control: Implement the join in a correctly partitioned AWS Glue Spark ETL job. The intended outcome is that large transformations can execute in distributed workers rather than one memory-limited process. The important limitation is that poor partition keys and skew can still dominate Spark runtime.

Incorrect Answers

This action is appropriate for another requirement: Transaction dollar amounts span several orders of magnitude, and a linear model is dominated by the largest outliers despite consistent labels. It does not solve the presented constraint: A 4 TB training dataset needs joins across multiple S3 partitions, and a single-node pandas job is exceeding memory. The matching action for this case is: Implement the join in a correctly partitioned AWS Glue Spark ETL job.

This action is appropriate for another requirement: Analysts need to compare transformations, visualize feature distributions, and export reproducible SageMaker processing code without maintaining a bespoke notebook transformation library. It does not solve the presented constraint: A 4 TB training dataset needs joins across multiple S3 partitions, and a single-node pandas job is exceeding memory. The matching action for this case is: Implement the join in a correctly partitioned AWS Glue Spark ETL job.

This action is appropriate for another requirement: A service classification model receives free-text customer tickets and must convert words into consistent model input tokens while preserving the same vocabulary during inference. It does not solve the presented constraint: A 4 TB training dataset needs joins across multiple S3 partitions, and a single-node pandas job is exceeding memory. The matching action for this case is: Implement the join in a correctly partitioned AWS Glue Spark ETL job.

Question 64

A cloud ML change request describes the following issue: Analysts need to compare transformations, visualize feature distributions, and export reproducible SageMaker processing code without maintaining a bespoke notebook transformation library. Which action best addresses the actual requirement?

Perform timestamp-aware historical feature joins using available offline Feature Store records.
Set up a Ground Truth labeling job with quality controls and appropriate workers.
Implement the join in a correctly partitioned AWS Glue Spark ETL job.
Create a versioned Data Wrangler flow and export its supported transformations into the ML pipeline.

Explanation

Correct Answer

Incorrect Answers

This action is appropriate for another requirement: A model that predicts defaults at loan origination currently joins the customer's current account balance rather than the balance known on the origination date. It does not solve the presented constraint: Analysts need to compare transformations, visualize feature distributions, and export reproducible SageMaker processing code without maintaining a bespoke notebook transformation library. The matching action for this case is: Create a versioned Data Wrangler flow and export its supported transformations into the ML pipeline.

This action is appropriate for another requirement: A retailer with an existing SageMaker Ground Truth account has thousands of ambiguous product images and needs human labels with periodic consensus verification for supervised classification. It does not solve the presented constraint: Analysts need to compare transformations, visualize feature distributions, and export reproducible SageMaker processing code without maintaining a bespoke notebook transformation library. The matching action for this case is: Create a versioned Data Wrangler flow and export its supported transformations into the ML pipeline.

This action is appropriate for another requirement: A 4 TB training dataset needs joins across multiple S3 partitions, and a single-node pandas job is exceeding memory. It does not solve the presented constraint: Analysts need to compare transformations, visualize feature distributions, and export reproducible SageMaker processing code without maintaining a bespoke notebook transformation library. The matching action for this case is: Create a versioned Data Wrangler flow and export its supported transformations into the ML pipeline.

Question 65

A team is choosing a specific control for this workload: A data operations analyst cleans inconsistent dates and malformed postal codes in a structured dataset using an auditable sequence of low-code rules. Which action best addresses the actual requirement?

Apply an appropriate log1p-style transform to nonnegative transaction amounts before training.
Use a training-fitted one-hot encoding transform with a defined unseen-category policy.
Perform timestamp-aware historical feature joins using available offline Feature Store records.
Create and execute a versioned AWS Glue DataBrew recipe on the incoming dataset.

Explanation

Correct Answer

Incorrect Answers

This action is appropriate for another requirement: Transaction dollar amounts span several orders of magnitude, and a linear model is dominated by the largest outliers despite consistent labels. It does not solve the presented constraint: A data operations analyst cleans inconsistent dates and malformed postal codes in a structured dataset using an auditable sequence of low-code rules. The matching action for this case is: Create and execute a versioned AWS Glue DataBrew recipe on the incoming dataset.

This action is appropriate for another requirement: A small product category field has red, blue, and green values without any intrinsic order, but an engineer currently maps them to 1, 2, and 3. It does not solve the presented constraint: A data operations analyst cleans inconsistent dates and malformed postal codes in a structured dataset using an auditable sequence of low-code rules. The matching action for this case is: Create and execute a versioned AWS Glue DataBrew recipe on the incoming dataset.

This action is appropriate for another requirement: A model that predicts defaults at loan origination currently joins the customer's current account balance rather than the balance known on the origination date. It does not solve the presented constraint: A data operations analyst cleans inconsistent dates and malformed postal codes in a structured dataset using an auditable sequence of low-code rules. The matching action for this case is: Create and execute a versioned AWS Glue DataBrew recipe on the incoming dataset.

Question 66

An ML engineering team faces this production constraint: A logistics stream requires stateful rolling five-minute counts from Kinesis records without launching and managing an always-on EC2 Spark cluster. Which action best addresses the actual requirement?

Fit a median imputer to training data and apply it consistently to subsequent data splits.
Process the Kinesis stream with an appropriate Managed Service for Apache Flink application.
Set up a Ground Truth labeling job with quality controls and appropriate workers.
Apply an appropriate log1p-style transform to nonnegative transaction amounts before training.

Explanation

Correct Answer

This action fits the situation: A logistics stream requires stateful rolling five-minute counts from Kinesis records without launching and managing an always-on EC2 Spark cluster. It implements the needed technical control: Process the Kinesis stream with an appropriate Managed Service for Apache Flink application. The intended outcome is that windowed aggregate features update continuously with supported state management. The important limitation is that incorrect event-time handling or checkpoint failures can produce late or duplicated aggregates.

Incorrect Answers

This action is appropriate for another requirement: A claims table has numeric repair costs with a long right tail and a few missing entries, making mean replacement highly sensitive to large claims. It does not solve the presented constraint: A logistics stream requires stateful rolling five-minute counts from Kinesis records without launching and managing an always-on EC2 Spark cluster. The matching action for this case is: Process the Kinesis stream with an appropriate Managed Service for Apache Flink application.

This action is appropriate for another requirement: A retailer with an existing SageMaker Ground Truth account has thousands of ambiguous product images and needs human labels with periodic consensus verification for supervised classification. It does not solve the presented constraint: A logistics stream requires stateful rolling five-minute counts from Kinesis records without launching and managing an always-on EC2 Spark cluster. The matching action for this case is: Process the Kinesis stream with an appropriate Managed Service for Apache Flink application.

This action is appropriate for another requirement: Transaction dollar amounts span several orders of magnitude, and a linear model is dominated by the largest outliers despite consistent labels. It does not solve the presented constraint: A logistics stream requires stateful rolling five-minute counts from Kinesis records without launching and managing an always-on EC2 Spark cluster. The matching action for this case is: Process the Kinesis stream with an appropriate Managed Service for Apache Flink application.

Question 67

A cloud ML change request describes the following issue: A retailer with an existing SageMaker Ground Truth account has thousands of ambiguous product images and needs human labels with periodic consensus verification for supervised classification. Which action best addresses the actual requirement?

Encode the ordered tiers using a deliberate ordinal mapping and handle unknown tiers explicitly.
Set up a Ground Truth labeling job with quality controls and appropriate workers.
Apply an appropriate log1p-style transform to nonnegative transaction amounts before training.
Fit the standardization parameters on training data only and apply the frozen transformation to validation and test splits.

Explanation

Correct Answer

This action fits the situation: A retailer with an existing SageMaker Ground Truth account has thousands of ambiguous product images and needs human labels with periodic consensus verification for supervised classification. It implements the needed technical control: Set up a Ground Truth labeling job with quality controls and appropriate workers. The intended outcome is that a reviewed labeled dataset becomes available for model development. The important limitation is that worker agreement does not prove labels are unbiased; Ground Truth ceased onboarding new customers on July 30, 2026.

Incorrect Answers

This action is appropriate for another requirement: A risk rating has meaningful ordered tiers low, medium, and high, and a product scoring model should retain that order in a compact feature. It does not solve the presented constraint: A retailer with an existing SageMaker Ground Truth account has thousands of ambiguous product images and needs human labels with periodic consensus verification for supervised classification. The matching action for this case is: Set up a Ground Truth labeling job with quality controls and appropriate workers.

This action is appropriate for another requirement: Transaction dollar amounts span several orders of magnitude, and a linear model is dominated by the largest outliers despite consistent labels. It does not solve the presented constraint: A retailer with an existing SageMaker Ground Truth account has thousands of ambiguous product images and needs human labels with periodic consensus verification for supervised classification. The matching action for this case is: Set up a Ground Truth labeling job with quality controls and appropriate workers.

This action is appropriate for another requirement: A churn model uses numerical features with different units, but the team accidentally fits a scaling transform on the complete dataset before holding out validation records. It does not solve the presented constraint: A retailer with an existing SageMaker Ground Truth account has thousands of ambiguous product images and needs human labels with periodic consensus verification for supervised classification. The matching action for this case is: Set up a Ground Truth labeling job with quality controls and appropriate workers.

Question 68

A team is choosing a specific control for this workload: A model that predicts defaults at loan origination currently joins the customer's current account balance rather than the balance known on the origination date. Which action best addresses the actual requirement?

Apply an appropriate log1p-style transform to nonnegative transaction amounts before training.
Create a versioned Data Wrangler flow and export its supported transformations into the ML pipeline.
Perform timestamp-aware historical feature joins using available offline Feature Store records.
Create explicit numeric bins with documented boundaries and a consistent unseen-value policy.

Explanation

Correct Answer

This action fits the situation: A model that predicts defaults at loan origination currently joins the customer's current account balance rather than the balance known on the origination date. It implements the needed technical control: Perform timestamp-aware historical feature joins using available offline Feature Store records. The intended outcome is that training rows contain only feature values that existed when each prediction would have been made. The important limitation is that a simple latest-value join can yield impressive but invalid offline accuracy.

Incorrect Answers

This action is appropriate for another requirement: Transaction dollar amounts span several orders of magnitude, and a linear model is dominated by the largest outliers despite consistent labels. It does not solve the presented constraint: A model that predicts defaults at loan origination currently joins the customer's current account balance rather than the balance known on the origination date. The matching action for this case is: Perform timestamp-aware historical feature joins using available offline Feature Store records.

This action is appropriate for another requirement: Analysts need to compare transformations, visualize feature distributions, and export reproducible SageMaker processing code without maintaining a bespoke notebook transformation library. It does not solve the presented constraint: A model that predicts defaults at loan origination currently joins the customer's current account balance rather than the balance known on the origination date. The matching action for this case is: Perform timestamp-aware historical feature joins using available offline Feature Store records.

This action is appropriate for another requirement: A credit risk team wants a smooth age measurement grouped into stable policy bands rather than a continuously varying predictor for a regulated report. It does not solve the presented constraint: A model that predicts defaults at loan origination currently joins the customer's current account balance rather than the balance known on the origination date. The matching action for this case is: Perform timestamp-aware historical feature joins using available offline Feature Store records.

Question 69

An ML engineering team faces this production constraint: A team needs to run a one-off Python preprocessing script using controlled container libraries and write transformed data to S3 without a long-lived notebook. Which action best addresses the actual requirement?

Run the preprocessing logic as a SageMaker Processing job with versioned code and output paths.
Use a training-fitted one-hot encoding transform with a defined unseen-category policy.
Apply an appropriate log1p-style transform to nonnegative transaction amounts before training.
Fit a median imputer to training data and apply it consistently to subsequent data splits.

Explanation

Correct Answer

This action fits the situation: A team needs to run a one-off Python preprocessing script using controlled container libraries and write transformed data to S3 without a long-lived notebook. It implements the needed technical control: Run the preprocessing logic as a SageMaker Processing job with versioned code and output paths. The intended outcome is that transformations execute as a reproducible managed job with bounded resources. The important limitation is that a Processing job does not itself automatically deploy or train a prediction endpoint.

Incorrect Answers

This action is appropriate for another requirement: A small product category field has red, blue, and green values without any intrinsic order, but an engineer currently maps them to 1, 2, and 3. It does not solve the presented constraint: A team needs to run a one-off Python preprocessing script using controlled container libraries and write transformed data to S3 without a long-lived notebook. The matching action for this case is: Run the preprocessing logic as a SageMaker Processing job with versioned code and output paths.

This action is appropriate for another requirement: Transaction dollar amounts span several orders of magnitude, and a linear model is dominated by the largest outliers despite consistent labels. It does not solve the presented constraint: A team needs to run a one-off Python preprocessing script using controlled container libraries and write transformed data to S3 without a long-lived notebook. The matching action for this case is: Run the preprocessing logic as a SageMaker Processing job with versioned code and output paths.

This action is appropriate for another requirement: A claims table has numeric repair costs with a long right tail and a few missing entries, making mean replacement highly sensitive to large claims. It does not solve the presented constraint: A team needs to run a one-off Python preprocessing script using controlled container libraries and write transformed data to S3 without a long-lived notebook. The matching action for this case is: Run the preprocessing logic as a SageMaker Processing job with versioned code and output paths.

Question 70

A cloud ML change request describes the following issue: A service classification model receives free-text customer tickets and must convert words into consistent model input tokens while preserving the same vocabulary during inference. Which action best addresses the actual requirement?

Use a training-fitted one-hot encoding transform with a defined unseen-category policy.
Encode the ordered tiers using a deliberate ordinal mapping and handle unknown tiers explicitly.
Fit or choose a supported text tokenizer and reuse its saved vocabulary and preprocessing settings for inference.
Implement the join in a correctly partitioned AWS Glue Spark ETL job.

Explanation

Correct Answer

This action fits the situation: A service classification model receives free-text customer tickets and must convert words into consistent model input tokens while preserving the same vocabulary during inference. It implements the needed technical control: Fit or choose a supported text tokenizer and reuse its saved vocabulary and preprocessing settings for inference. The intended outcome is that text inputs are converted into consistent token IDs accepted by the model. The important limitation is that changing tokenization between training and serving can invalidate learned embeddings.

Incorrect Answers

This action is appropriate for another requirement: A small product category field has red, blue, and green values without any intrinsic order, but an engineer currently maps them to 1, 2, and 3. It does not solve the presented constraint: A service classification model receives free-text customer tickets and must convert words into consistent model input tokens while preserving the same vocabulary during inference. The matching action for this case is: Fit or choose a supported text tokenizer and reuse its saved vocabulary and preprocessing settings for inference.

This action is appropriate for another requirement: A risk rating has meaningful ordered tiers low, medium, and high, and a product scoring model should retain that order in a compact feature. It does not solve the presented constraint: A service classification model receives free-text customer tickets and must convert words into consistent model input tokens while preserving the same vocabulary during inference. The matching action for this case is: Fit or choose a supported text tokenizer and reuse its saved vocabulary and preprocessing settings for inference.

This action is appropriate for another requirement: A 4 TB training dataset needs joins across multiple S3 partitions, and a single-node pandas job is exceeding memory. It does not solve the presented constraint: A service classification model receives free-text customer tickets and must convert words into consistent model input tokens while preserving the same vocabulary during inference. The matching action for this case is: Fit or choose a supported text tokenizer and reuse its saved vocabulary and preprocessing settings for inference.

Amazon AWS Certified Machine Learning Engineer - Associate MLA-C01 Exam Dumps, Practice Test Questions

Question 60

Explanation

Correct Answers

Incorrect Answers

Question 61

Explanation

Correct Answers

Incorrect Answers

Question 62

Explanation

Correct Answer

Incorrect Answers

Question 63

Explanation

Correct Answer

Incorrect Answers

Question 64

Explanation

Correct Answer

Incorrect Answers

Question 65

Explanation

Correct Answer

Incorrect Answers

Question 66

Explanation

Correct Answer

Incorrect Answers

Question 67

Explanation

Correct Answer

Incorrect Answers

Question 68

Explanation

Correct Answer

Incorrect Answers

Question 69

Explanation

Correct Answer

Incorrect Answers

Question 70

Explanation

Correct Answer

Incorrect Answers